ML25168A013

11-06-84 ACRS Reply to Questions Regarding the Proposed NRC Severe Accident Policy Statement

ML25168A013
Person / Time
Issue date:	11/06/1984
From:	Ward D Advisory Committee on Reactor Safeguards
To:	Asselstine J NRC/OCM
References
Download: ML25168A013 (1)

Text

UNITED STATES NUCLEAR REGULATORY COMMISSION ADVISORY COMMITTEE ON REACTOR SAFEGUARDS WASHINGTON, D. C. 20555 Honorable James K. Asselstine Commissioner November 6, 1984 U.S. Nuclear Regulatory Commission Washington, D.C.

20555

SUBJECT:

ACRS REPLY TO QUESTIONS REGARDING THE PROPOSED NRC SEVERE ACCIDENT POLICY STATEMENT This is in reply to your memorandum of September 6, 1984 (a copy of which is attached).

Your memorandum and the Committee report dated July 18, 1984 on NUREG-1070 were both addressed to the draft version of the Severe Accident Policy Statement dated April 18, 1984.

Some of your concerns were with questions similar to some of those raised in the comments and recommendations in the Committee report, including its additional comments.

Subsequently, in the proposed Federal Register Notice, Policy Statement, and the draft of NUREG-1070 dated August 23, 1984, a number of significant revisions were made to the earlier ver-sions.

Those revisions go far toward meeting some of the concerns raised in your letter, as well as in the Committee report.

In response to your particular questions, we have the following com-ments:

1.

You ask for a qualitative judgment regarding the likelihood of an accident resulting in serious core damage before the end of the twen-tieth century.

As you pointed out, a precise numerical estimate is not possible; but at the same time, the operational utility of any sub-jectively generated phrase such as, "somewhat unlikely" or "not un-likely" is limited.

Some admittedly simplistic observations may, nevertheless, be made. If one merely uses the statement in the present proposed Quantitative Safety Goal (fewer than one in 10,000 incidents per reactor year) and considers only nuclear power plants in the U.S., then, with.-vlOO plants over a period of 15 years, one comes up with the probability estimate of 0.15. It is not really known, of course, whether the present population of plants meets this goal: by some current estimates, some existing p 1 ants fail to meet it by a factor of 3 or more, whi 1 e others exceed it by a similar factor.

In addition, by some estimates, the occurrence frequency of earthquakes beyond the seismic design basis is, in some cases, as large or larger than one in 10,000 per year.

Such seismic hazard estimates are accompanied by very great differences of opinion among experts.

Moreover, the general matter of component fragility inc 1 udes many i terns on which subjective judgment provides the only 1985

Honorable James November 6, 1984 guidance presently available.

Although there is reason to expect that some nuclear power plants can ride out an earthquake considerably larger than the seismic design basis, not all plants have yet been carefully examined on this point.

We don't really know exactly how the present population stands with respect to the safety goal.

Two things must be kept in mind concerning both the general probability estimate of 0.15 and the more specific estimates of the probable fre-quency of some event or other obtained from probabilistic risk assessments {PRAs).

On the one hand, such quantities should on no account be taken as representing an actual value, while on the other, they should not be completely ignored.

The frequently reiterated attention called to the uncertainty bands associated with such numbers should be warning enough against assigning them literal significance.

(For this reason, for example, their direct use in cost/benefit analyses must be treated with great caution.

As a further aside, it may be worth noting that if only one quarter of the population of reactors should have, in fact, a probab 1 e frequency for core-damaging accidents four times that proposed for the Safety Goal -- a situation within the realm of possibility -- then the probability of a serious accident within this subgroup alone would equal 0.15, quite irrespective of the relative superiority of the rest of the population.)

However, despite the necessarily fuzzy nature of the sort of numbers referred to, they ought not to be ignored, since they are capable of giving some indication of the state of affairs and, in many cases, the most useful indication available.

Whatever the "true" value of the probability in question may be (i.e.,

whether larger or smaller at this moment than the 0.15 arbitrarily mentioned above), it should be recognized that this quantity has been reduced considerably over the past five years and is 1 i kely to move toward lower values in the coming years.

Without attempting to quantify this progress s it is the certain consequence of improvements in such things as operator training, plant design, instrumentation, procedures, simulator experience, and so forth, as we 11 as of specific fixes.

Of similar importance, since TMI, are the factors of increased awareness and attention to safety on the part of 1 icensee management and the nuclear industry generally.

Although most of the changes have been within the spectrum of design basis events and are not yet appreciably affected by specific consideration of severe accident problems, many severe accidents (e.g., TMI) are merely advanced stages of something which started as a design basis event.

Continuing improvement in the actual accident frequency value for the reactor pcpulation will result from the programs addressed to Unresolved Safety Issues (USis), generic issues, lessons from operational experience, and, perhaps particularly, from the correction of any major defects which may be identified in the planned plant-specific survey.

All this is in addition to whatever may emerge from the Severe Accident Research Program.

Only some fraction of core-damaging accidents is expected to result in serious offsite consequences.

This fraction is also not really known, though it has sometimes been pictured as being about one-tenth.

The studies of containment performance may reduce this fraction over the coming years.

The Source Term Program will affect the points considered 1986

Honorable James November 6, 1984 here, since it is important to estimates of the severity of consequences of such accidents as may occur.

In conclusion, and as an essentially intuitive judgment, we would consider it reascnable to suppose that, with our present {and proposed) program and a population of,-_,100 reactors, the probability of a core-degrading accident between now and the end of the century may lie in the neighborhood of one in ten -- within a factor of a few either way.

Whether this should be phrased as "unlikely," or "not highly unlikely," or just how, we must leave to you.

2.

Whether the occurrence of a severe accident within the next 16 years will "have very serious adverse consequences on public confidence" would probably depend {a) on the timing, (b) on the extent to which public opinion makers and policy makers can and do mature in their understand-ing of the risk from nuclear power, especially in its relationship to risk in general, and {c) on the nation's perceived need for nuclear energy.

While we have no special expertise in evaluating anticipated public reactions, we would believe that another TMI-type accident in 1985 would have a strong adverse effect, whereas one in 1995 would probably cause much less reaction.

While the Corrvnission should take responsible actions to assure the safety of nuclear power plants, it should limit its actions to those things required to provide an appropriate balance between risk and cost.

The Commission should take the responsibility for determining how safe plants need to be and for evaluating the performance of plants to assure that they operate at or above that minimum level of safety. The Convnis-sion should not require expenditures of ratepayers', investors', or taxpayers' resources solely to improve the public perception of the safety of nuclear power.

That responsibility must lie with the nuclear utility industry, or if it becomes a question of national policy, with the Congress.

The Commission should assure itself that it has the necessary resources to fulfill its respor.sibility to assure an appropri-ate level of safety in nuclear power generation and take a clear, steady path to that goal.

The Commission's main responsibility with respect to public perception is to establish and maintain a strong reputation as a responsible, capable, independent regulatory body.

It should not accept industry proposals without its own careful review and evaluation.

Nor should it allow the impression, by acceding to unfounded criticism, that nuclear power risks are higher than they really are.

It should, of course, l is ten to a 11.

We do not suppose that the matters of public confidence or public perception should be specifically discussed in the Pol icy Statement (although it is alluded to at the end of Section B.2 of the Policy Statement), but the Commission cannot escape having to take these views into account in all its activities.

However, the intent of the Corrmis-sion to "take all reasonable steps" (as spelled out in your "objective" statement) should be unmistakably clear.

We have no doubt that this is the intent, but it does have to be read into the Policy Statement.

Even though the presence or absence of some explicit phrase expressing the 1987

Honorable James November 6, 1984 intent might not significantly affect what may actually happen, it would serve the purpose of clarification and providing perspective.

We believe that such an expression would improve the Policy Statement.

As to where it might most appropriately appear, it would seem possible, since the second sentence of the Introduction is somewhat out of context in the present text of the Policy Statement, that the Introduction could start with this second sentence and be followed by something similar to your phrasing of the overall intent.

This should be followed by a second paragraph discussing the documentation aspect of things --

as presently laid out.

3.

The occurrence of a severe accident is, of course, more likely at some plants than others.

In some cases, this may be attributable to deficiencies in systems or layout or procedures:

things which may, for example, be identified through the proposed plant-by-plant survey.

In all cases, management commitment and attention to safety are critical.

The quality of operation is not the same at all plants.

The ability of operators, the quality of equipment maintenance and testing programs, the comprehensiveness of technical support, and the overall ability of plant management are important considerations in operation.

There does not exist, at present, satisfactory means for fully evaluating these factors.

The NRC should give high priority to programs that will improve, or encourage the utilities to improve, the quality of opera-tion.

The Institute for Nuclear Power Operations (INPO) has, a~d should continue to have, a key role in this area.

But, it must be realized that the INPO role will be in assisting the utilities toward improving quality and there will remain a responsibility for the NRC to monitor and evaluate performance.

This probably cannot be done effectively by strict and prescriptive methods, as can be done for hardwc1re systems, but will require a fresh approach.

Development of an effective and practical approach should be a high-priority matter for the NRC.

4.

You asked what further steps could be taken.

Our letter of July 18, 1984 rec0mmended that "an appropriate approach" be developed for syste-matic analysis of each plant not previously analyzed.

We do not recom-mend a full-scale PRA for each plant.

So far as "focusing attention on weak performers" is concerned, if this means publishing periodically a list of rankings of weakness -- of course not.

If it means having the NRC Staff direct special attention to points where it has perceived s pee i a 1 prob 1 ems --

by c111 means.

The NRC Sta ff is not in a good position to assess weak management, or to do suitably constructive things about it when it is identified.

This is an area where the INPO has a strong interest, as well as being in a much better position to be effective.

A second broad step, relating to potential vulnerabilities or inade-quacies in design, is under way by such work as is being conducted under USI A-45, "Shutdown Decay Heat Removal Requirements." This work appears to have possibilities for acceleration and that would be desirable.

"Jl.ssured core cooling" is a keystone goal of all regulatory activities, and it would seem that design improvement in some specific plants or nuclear steam supply systems may be desirable.

For example, considera-tion might be given to possible upgrading of some of the General Design Criteria, such as Criterion 34 - Residual Heat Removal.

The question of 1988

Honorable James November 6, 1984 the relative merits of redundancy in components versus diversity in design is complex and worthy of attention.

This is a further step, which would require time, thought, and effort to accomplish.

The promotion and assurance of safety are the principal reasons for NRC's regulatory activities including inspection, enforce-ment, etc.

Though not its only objective, the achievement and mainten-ance of safety is equally essential to the nuclear industry.

The full recognition of this throughout the NRC Staff and all branches of the industry, along with such care in the conduct of NRC's activities that it becomes apparent to those others affected that this is the compelling end in view, could provide the basis for a common, cooperative, and even enthusiastic commitment to safety, and to excellence in performance on which safety finally depends.

This step, or even progress in this direction, would be more important than any of those spelled out in the Policy Statement, or elsewhere in this letter.

5.

We believe the present Policy Statement does not overemphasize the use of cost-benefit analyses and PRA.

It does not, and we do not, recommend that the results of either of these types of analyses be used to yield automatic decisions based on numerical answers.

Both cost-benefit analyses and PRA are useful tools that may furnish necessary inputs to engineering judgment, but as the ACRS noted in its letter of July 18, 1984, prudence and sophistication will be required in their use.

We question whether there is any such thing as qualitative engineering judgment.

Engineering is by its nature a quantitative art and the issues to be addressed by a severe accident policy cannot escape having a major quantitative component.

Until something better comes along, cost-benefit analysis and PRA are the best means for addressing these inherently quantitative questions.

The important uncertainties in analytical results are implicit in the issues, and are not created by the methods of analysis.

6.

You asked if the possible need for major changes should be made clearer. This point has been much improved in the present versions: it appears in Paragraph 3 of the Federal Register and in Paragraph 1 and Section C.1, Paragraph 3, of the Policy Statement.

We recommend further changes.

For example, the wording used in the "Commission concludes" sentence in the Federal Register would be preferable to that used in the analogous sentence in the Policy Statement, Section C, where the word "dismisses" has an unfortunate air.

It would also be helpful (in these same two locations) if the whole content of the "ongoing programs, USis, etc.," paragraph were to follow immediately (in the same paragraph) after the 11Commission concludes 11 sentence; while the whole discussion concerning needs for and plans for plant-specific surveys was put in a following paragraph.

This would strengthen the point that if major changes appear to be needed the Commission will require them.

7.

Resolution of USI A-45 is important in consideration of a Severe Accident Policy in that it could be one of the most significant programs the Convnission has under way to reduce the likelihood of occurrence of a severe accident.

The present A-45 program is, as you suggest, 11a broad 1989

Honorable James November 6, 1984 based program.... 11 The resolution of A-45 may permit, however, "only...

procedural changes to improve the reliability of existing decay heat removal systems... " in cases where only that sort of improvement 1s necessary.

There is a broad range in reliability and capability of existing decay heat removal systems and the improvements required will account for that. Backfits are likely to be quite plant-specific.

Additional comments by ACRS Member Harold Lewis, Member David Okrent, and Members Harold Lewis, Dade Moeller, and David Ward are presented below.

Sincerely,

~-0O.~Q David A. Ward Acting Chairman Additional Comments by Jl.CRS Member Harold Lewis This is a generally excellent letter.

I only wish to emphasize that uncertainties in quantitative estimates of probability go in both directions, and that there are those who believe that seismic risk, singled out in this letter for its large uncertainty, is a case of risk overestimation.

Many of the PRAs used in our estimates were performed in a regulatory climate, for a regulatory purpose, and are infected with unquantified conservatisms.

PRAs must be done rea listica 1 ly to earn credible use in a predictive role.

All of our numbers should be read with this in mind.

Additional Comments by ACRS Member David Okrent In its answer to question 1, the ACRS has been careful to identify the large uncertainties inherent in any estimate of average core melt frequency for the entire population of U.S. reactors.

However, in view c,f factors such as the cor.siderable number of plants designed before 1970, the frequency with which the NRC best estimate of core melt frequency for existing full-scale PRAs exceeds one in ten thousand per reactor year, and the fact that even the better PRAs are incomplete, my own intuitive judgment is somewhat more pessimistic than that given by the ACRS, say by a factor of about 3.

I would say that a core melt accident by the year 2000 is not unlikely.

Additional Comrrents by ACRS Members Harold Lewis, Dade Moeller, and David Ward If the Conmission becomes preoccupied with the issue of public per-ception of safety, it might shirk its responsibility to lead and to make definitive judgments about nuclear power safety. It has ready access to talents and resources about nuclear safety that the public does not, and that the Congress does not.

It should use those resources.

While the 1990

Honorable James November 6, 1984 Congress and the public properly have the right and power to make ultimate decisions, neither body has the ability to provide expert leadership. That is what the Commission must do.

Attachment:

Memorandum from James K. Asselstine, NRC Commissioner, to Jesse C. Ebersole, ACRS Chairman, "Severe Accident Policy Statement," dated Sept. 6, 1984 1991